10th May 2006 Hoyle committee room 13.00 - 15:15 Present: DWE, MJI, JRL, PSB, STH, MR, RGM, WJS (to 14:00) Apologies: NAW, EGS Agenda ------ 1. Actions from last meeting 2. Comments on WFAU minutes 3. Report on VDMT meeting 4. Data archives update 5. Optical processing update 6. WFCAM news 7. VDFS news 8. AstroGrid liaison update 9. AOB Minutes ------- 1. Actions from last meeting JRL has looked into how to transfer the MSB quality information into the processed files. The problem is that there appears to be no simple way to link the OMP frame identifier to an MSB frame sequence. The entries in the database are manually entered so are variable in content. Since this cannot be resolved, it has been decided to drop this action. MJI has made no further progress investigating the merging of classification statistics - ongoing <<<< MJI raised the issue of the CASU preferred method of data transfer from ESO at the VDMT meeting. Our preference is for LTO tapes. MJI noted that ESO are planning to use LTO tapes as part of their upgraded tape archive system. WJS has contacted JPE regarding acquiring the numeric (rather than graphic) versions of the data that is required by the ETC. However, no real progress has been made in acquiring any data. MJI wondered if we could get an official VPO digitization of the graphs. WJS thought that this would be unlikely considering how busy the VPO were. It was agreed that what we need is for VPO to definitively say that we will not get any data in digital form and then PSB could be assigned to organise digitizing the graphs in detail and produce the best available information. WJS said that he would contact VPO to clarify this. <<<< RGM is working on the foul-up causes in LAS outliers list. A report is being written. He gave a summary of the main causes: deblending problems - a software bug has been fixed as discussed previously. But seeing variations are the main cause. RGM requested that a deblending flag be placed in the catalogue. MJI replied that it was already flagged in the areal profile; edge effects - most numerous cause. Can this be flagged? MJI thought that a simple definition was problematic (variable aperture size, odd detector properties near the boundary) though a prototype "no. of bad pixels" flag was already under test. RGM said that he had found that an empirical edge limit of 32 pixels worked satisfactorily; multicolour orphans - this affected around 0.1% of sources in the LAS EDR and appears to be a feature of the matching algorithm at the archive end. It manifests itself in the LAS as cases where an isolated unblended source detected in multiple wavebands, is not being bandmerged correctly even though all the individual detections are within a common area with radius 1". It appears as two sources with peculiar colours. This feature has been reported to Nigel Hambly. crosstalk artefacts - these make up ~1% of objects but should only be a problem in the EDR version 05A data. RGM noted that he could not test 05B data, with the cross-talk problem fixed, since it was not available; moving objects - solar system objects can cause problems but this is often readily detectable from the movement across the filters within an MSB; RGM also pointed out that the saturation flag was not set in the archive. MJI said that this was correct and that the CASU documentation is wrong here, since the CASU matching software computes and updates this flag not the catalogue generating software. MJI said that he would email Nigel Hambly warning him of this. <<<< STH finally asked Floor van Leeuwen what had happened at the AstroGrid Implementation meeting - answer = nothing of note. ALL sent in reports to STH covering the entire quarter. DWE placed a link in the persistence report to the relevant 2MASS persistence web page JRL think they have fixed the remaining reported header problems for the raw MJI 05A data. These were linked to jitter sequence keywords. It was noted that this work still continues as new header problems in PI data are discovered during reprocessing. MR have finished checking the test sets of reprocessed data from 05A for MJI quality control but note that they continue to check the data for QC as it is reprocessed, which is standard CASU practice. MJI distributed an updated Q1/Q2 plan RGM did decide that his telecon presence at the VDMT meeting was wanted, but this did not happen due to a misunderstanding. JRL has investigated the UKIRT archive missing header information problem. A solution has been found but not implemented yet - ongoing <<<< WJS has passed on the request to RAL that CASU want whole data frames not just cutouts. PSB confirmed that recent data received were whole frames. PSB wrote some code to reorder the extensions for AIT FITS files 2. Comments on WFAU minutes PSB reported on the current round of transfer tests. A couple of days after the VDMT meeting where this was discussed, the high performance SSH package (http://www.psc.edu/networking/projects/hpn-ssh/) was installed at the IoA. We are now waiting for the appropriate changes to be done at the WFAU end so that further tests can be carried out. This may make transfers faster and ameliorate the Leeds bottleneck. The relevant information regarding UKLight has been fed back to Johann Bryant so that the UKLight form and science case can be filled in and circulated for comment. MJI reemphasised that the main reason CASU are in favour of UKLight was because the "running" costs will be less than the present ridiculous (£1k/TB). It was reported that the number of threads used for network transfer from apm7 had been adjusted in response to our request and that the performance of the machine for other users was noticeably better. STH noted the release schedule of DR1 in the minutes and commented on the fraction of PATT PI data which had been released as flat files through the archive. 3. Report on VDMT meeting MJI, STH and PSB attended the 11 April VDMT meeting in Edinburgh and commented that there was much discussion regarding UKLight and data transfers. There was also some discussion regarding possible release of the processed flat data files for UKIDSS (in a similar way as access to PATT PI data is already provided). The UK review for VDMT will take place in October and JPE is attempting to round up panel members. 4. Data archives update WFCAM: MR reported that the number of requests for raw WFCAM data through the ESO archive was negligible. UKIRT: A missing header fix is about to be implemented (see earlier comment). ING: Meetings between EGS and JRL have been held to organize bringing the archive up-to-date. At the moment the online database is only current up to 2002 (although data up to the present is available on disc). The plan is to build the database with a selected set of keywords, but also ingest the whole FITS header so that future expansion and upgrades can be readily accommodated. This should be completed by the end of the month. 5. Optical processing update MJI reported that not much has happened recently on this front. The main news is that all the 2005 data for the IPHAS (Galactic Plane Survey) has now been processed and released and that 75% of the survey has now been processed and is available online in standard pipeline product form, via QC database access and also as calibrated merged catalogue products. 6. WFCAM news Reprocessing 05A: MR said that up to 5 June had been, or is being, reprocessed and this is on track to be completed by mid-to-late May. No processing of 06A will be carried out until the reprocessing of 05A has been completed. However, as of yet, no 06A data has been received so this is not an issue. Upcoming UKIRT Board meeting: The UKIRT Board will meet at Hertford on 22/23 May and MJI will be attending and giving a short presentation. A report has already been submitted and will be made available on our internal website and Plone site. <<<< http://www.ast.cam.ac.uk/~wfcam/internal/meetings/ukirt/ 05B broken files: MJI reported that out of a total of more than 400,000 transferred files 7 (sic!) had been reported as "broken" during archive ingest:- 4 of these were engineering data that should never have been processed and hence had missing header data; 1 was a catalogue with overwritten columns - which was reprocessed and fixed as soon as it was reported; and 2 were down to transfer errors - 1 due to a power cut at ROE, the other an unknown FU during transfer. In both these latter cases the on-line versions here were fine. He noted that this is an amazingly high success rate for processing, transfer and ingest into an archive, and probably qualifies for the Guinness Book of Records. Hardware additions: All available WFCAM disk space was used up for 05A and 05B raw data and processed data storage. To accomodate 06A data MJI reported that one new RAID array has already been acquired, tested, setup and is already in use. This is a 16 slot fibre channel system (running in RAID 6 configuration with 7 Tbyte usable storage from the 16x500 Gbyte SATA disks) attached to one of the disk servers. He also noted that another system, is on order that will use similar technology but be based on the new Seagate 750 Gbyte SATA disks and will therefore give ~10 Tbyte of usable storage per unit. Costs are still about £1k/Tbyte but at least these systems are relatively simple to install. 2 new (actually second hand) 19inch racks have also been bought to house the RAID arrays and associated disk servers and a quote is being sought for installing a cooling unit in the "storeroom" to house the new kit. He also reported that APM12 will be used for data processing in addition to serving SybaseIQ, to give a bit more oomph at peak processing time. UKIDSS survey heads telecon: MJI reported on this telecon which was held last week. The main topic was data access policy. Cambridge astronomers are in favour of fast access for the flat files but surprisingly nobody else present seemed as keen. At the moment nothing will happen on this until DR1 has been released. WFAU will look into this problem after. MJI noted that as a result of the Ellis review a new call has been announced for major UKIRT programmes with letters of intent due early July. Photometry: STH said that he had contacted Roc Cutri regarding the 2MASS Touchstone fields. About 20 of these overlap WFCAM data. These are well-calibrated photometric fields which will help with WFCAM photometric calibration issues including the next topic. Non-linearity: STH gave a short presentation on the non-linearity investigations initiated by MJI . Comparisons with respect to 2MASS shows an effect at the 0.03 mag over 7.5 mags level. The effect is not a function of object colour, airmass nor seeing. There were some, possibly marginal, indications of a filter dependence. As a check on this JRL dug out some previously analysed April 2005 K-band dome flat sequences designed to investigate linearity. His provisional results contradicted what he had found previously. MJI was unhappy with this dome flat re-analysis and wanted further work to assess what was going on before making this more public. In particular the plots shown at the meeting were just the model fits to the data and he wanted to see the original data points before making a judgement. JRL was asked to provide graphs or tables showing the original data. <<<< ALL were keen to get something out soon on this, to describe the effect and indicate possible actions and consequences. The meeting also suggested that JRL ask for further dome sequences to be taken. <<<< STH was asked to produce some 2MASS comparison plots on a per detector basis for additional checks. <<<< Plone: DWE said that all files (public and internal) have now been copied to the Plone site and that the web site is regularly checked to see if files need to be copied across. The design of the site is ongoing and a further meeting of interested parties will be held sometime around 23-25 May when all are in Cambridge. A new idea has come up of creating a new Plone state of "obsolete". This will mean that we will have access to old versions of the documents, but that in most normal cases they will be invisible. This should lead to less confusion. STH asked about how easy it was to administer and alter the site. He was concerned about EGS being a single point failure for the Plone system. DWE felt that it was not an easy task to alter the site construction, although adding content was relatively easy. JRL asked DWE (& EGS) for help in setting up an ADASS 2007 web site since they are Plone-based <<<< 7. VDFS news VIRCAM Status AITnn: WJS reported on the status of the telescope, mirror, dome and camera. The camera all-up tests will start shortly and the camera is cooling down. WJS said that he had received a copy of the CASU wish list of test measurements. These tests will be carried out, if possible, through June. PSB and MJI reiterated their desire to participate in some of the tests if this is feasible. WJS said that Steven Beard had visited ESO and carried out some tests on the software on ESO hardware. The software now meets the write to disc requirement. The TCS and camera software has been altered so that operations can now be carried out in parallel, though whether ESO Operations entertain this is another matter. RGM and MJI pointed out that this was an important issue with respect to large area shallow survey overheads. There will be a meeting for VISTA Public Surveys in Edinburgh on 20/21 June and RGM suggested drawing up a table showing the benefits of parallelism. MJI asked who had taken over from Martin Folger working on the SADT. WJS noted that it was Nuria Lorente. VDFS docs status: PSB said that the DRLD draft was continuously being updated to correspond to the DRL software, but no progress had been made in releasing any of the outstanding documents. DRL status: JRL said that PSB is currently testing the software. The work is about 1 week behind our self-imposed schedule. The recipes, test data and associated documentation will be packaged up and released in a few days. <<<< ETC status: Nothing to report, but see earlier action comments. ESO meeting report: MJI reported that on 26 April he, JRL and PSB met at ESO with Michele Peron, Sandra Castro, Wolfgang Hummel and Pascal Ballester. The purpose of this meeting was to establish if a test DRL for VISTA could be delivered in time to be run as part of the ESO end-to-end dry-run system tests at Garching in June. MJI noted that ESO require the summit pipeline to do data quality tests on all the detectors and not just a random sub-sample. There will be a Blade (Athlon) cluster on Paranal, similar to the Garching system, to carry this out. ESO have a new archive system which is a combination of disc and (robotic) tape-based systems. Since this is an LTO system, this lends hope to the prospect of transferring data from ESO to CASU via LTO tape (Internet transfer seems to still be a non-starter). The intention is that ESO will buy the hardware for the summit pipeline and install the software at Garching prior to shipping, probably late Autumn. A CASU visit to Garching will probably be needed to expedite this. At the meeting JRL reported on the 7 recipes already written. A new recipe (microstep and jitter) was promised for the June tests. For the upcoming dry-run tests, it was noted that a mixture of real and simulated data would be preferable . However, the cross-talk and persistence recipes need laboratory data, and possibly on-sky commissioning data, to characterise the problems before they can be finished. It was further agreed that it was probably not necessary for a CASU presence at ESO for these tests. MJI said that it had been a fruitful meeting and good progress on several issues had been made. A copy of the minutes has been sent to JPE and JMS. 8. AstroGrid liaison update PSB mentioned that a bid for part of some local e-science money had gone in for the IoA to become part of CamGrid. He noted that it was unclear at this stage who would manage the local machines if this bid is successful. 9. AOB MJI pointed out the upcoming ESO Instrument Calibration Workshop (January 23-26 2007, Garching) http://www.eso.org/cal07 and encouraged people to think about attending. DWE brought the Cambridge Beer Festival to people's attention (22-27 May) http://www.cambridge-camra.org.uk/2006/cbf.html Continuing Actions ------------------ MJI finish investigating merged classification statistics JRL fix problem with UKIRT archive (missing header information problem) New Actions ----------- WJS contact VPO to confirm that we will get no more ETC-related digitized data from them MJI email Nigel Hambly regarding the non-implementation of the saturation flag in the classifier JRL report back to next meeting on progress with the ING archive update MJI place submitted UKIRT Board report on web site JRL provide non-linearity graphs or tables showing original data JRL request further dome sequences to be taken STH produce 2MASS comparison plots on a per detector basis DWE help JRL in setting up an ADASS 2007 web site EGS JRL package up and send out next DRL release